Base Command Manager / Bright Cluster Manager Release Notes
Release notes for Bright 7.2
== New Features ==
* Improved Puppet integration
** Puppet module management with PuppetForge
** fine grained puppet configuration
* Docker integration (allows containers to be started on nodes in a Bright cluster)
* Kubernetes integration (allows containers to be scheduled in a Bright cluster)
* ZFS support (for building storage nodes and as an alternative to btrfs for software image revision control)
* Basic Omnipath support (procedure for setting up and some basic metrics and healthchecks)
* Job based metrics
* DRAC bmc network interface support
* File Customizations to deal with additional configuration parameters
* UEFI support for head node installer
== OpenStack ==
* Upgrade OpenStack to Liberty
* Simplified OpenStack deployment process:
** by grouping OpenStack nodes into logical units (network, compute, controller) via Configuring Overlays
** streamlined the CMGUI OpenStack deployment wizard
* Added 'express' CMGUI wizard, which allows to deploy a default (reference) Bright OpenStack cloud, with minimum of questions to be answered
* Added ability to select preferred Keystone Authentication backend on deploy time (MySQL, LDAP, LDAP/AD, PAM)
* Added Open vSwitch support for Neutron ML2 (as an alternative to existing Linux Bridge support)
* Added builtin integration with Glance with Cinder Drivers:
** GPFS
** Solidfire
** 3PAR
** NetApp
** Dell Storage Center
* Added builtin integration with Glance Drivers:
** GPFS
* Bright OpenStack now comes with ELK stack (Elastic search, Logstach, Kibana)
* Bright OpenStack now comes with Galera for MariaDB, for scalable and fault tolerant database backend for OpenStack services.
* cm-openstack-setup featured a pluggable architecture, allowing users to easily extend Bright OpenStack with additional functionality
* Improved cm-openstack-setup deployment utility
* Restructured the OpenStack 'Settings' submode in cmsh and pythoncm
* Split the OpenStackNetwork role into
** OpenStackNetworkDHCPAgent
** OpenStackNetworkL3Agent
** OpenStackNetworkLBAgent
** OpenStackNetworkMetadata
** OpenStackNetworkOVSAgent
== Big Data ==
* Updated support for latest releases from Apache, Cloudera, Hortonworks, Pivotal:
** Apache Hadoop 2.7.1
** Cloudera CDH 5.3.8 and 5.4.8
** Hortonworks 2.2.8 and 2.3.2
** Pivotal HD 2.1.0 and 3.0.1
* Improved support for Apache Spark:
** added Spark wizard in CMGUI
** added support for Spark History Server
** added High Availability for Spark Master
** added more advanced configuration options
** deployment of Spark in Standalone Mode is now configurable via XML
** added support for HAL plugin. It allows Spark to access Lustre
** added Livy Job Server
* Improved support for Lustre and SLURM:
** added support for HAM plugin. It allows to run Hadoop jobs via SLURM and switch between YARN and SLURM
* Improved Apache Accumulo integration via dedicated Roles.
* Improved Apache Kafka integration via dedicated Role.
* Improved Apache Pig integration via dedicated Role.
* Improved Apache Storm integration via dedicated Roles.
* Added Apache Drill support via dedicated Role.
* Added Apache Flink support via dedicated Roles.
* Added Apache Zeppelin support via dedicated script.
* Added Apache Ignite support via dedicated script. It allows to switch between YARN and Ignite to execute MapReduce jobs.
* Added Tachyon support via dedicated script.
* Improved flexibility:
** JAVA_HOME can be changed after the deployment of a Big Data component.
** Hadoop can be deployed with both NameNode HA and YARN HA
== Monitoring ==
* Added MonConf on the level of Device, which allows collection metrics for a few devices without cluttering the other devices.
* Allow time intervals to be specified when actions should be executed
* Removed valid-for restrictions on metrics and health checks. Custom scripts which add metrics should remove the corresponding set statements.
== Changes in Behavior of Existing Features ==
* Configuration file customization for OpenStack/Hadoop/Ceph moved to Configuration overlays
* Improved IPv6 support for head node
* Added extra monitoring items to GPU metric collection script: PCI throughput, PCI replay count, encoder utilization, decoder utilization, number of procs using the GPU, BAR1 memory usage
* Improved DNS/named configuration writer to support zones that do are not /8 /16 /24
* Added Notes field to User for additional non ldap notes
* Speedup of DB read/write operations
* Improved threading inside cmdaemon
* Changed to new a /tftpboot structure for improved BIOS / EFI support
* Improved checking in pythoncm
* Added exceptions to pythoncm
** for get/set operations
** for remote procedure calls
* Default power control is now "none", and will be automatically changed to ipmi/ilo when an interface is created
* Added option to disable port detection per switch
* Slurm: for a systemd-enabled setup slurm service is now split into two native systemd services (slurmd and slurmctld) in addition to slurmdbd
== Big Data ==
* Added roles for:
** Apache Accumulo, Apache Kafka, Apache Pig, Apache Storm.
* An XML configuration file is now needed for the deployment of Spark in Standalone Mode.
== Miscellaneous Notes ==
* All packages were upgraded to latest versions